Math whiz Usama Fayyad, a NASA and Microsoft vet, is now prospecting with own startup, digiMine.
On a chilly Monday morning last March, Usama Fayyad met Nick Besbeas and Bassel Ojjeh for breakfast at their usual hangout, a restaurant in Kirkland, Wash., called Hector’s. The mood, recalls Besbeas, was a bit surreal. Days earlier, all three had ditched lucrative positions at Microsoft to start a company together, and reality was just sinking in.
Gone were their secure jobs, the talented staffs they had helped cultivate, the piles of stock options they has amassed during their Redmond careers. Looking across the table at his colleagues, Fayyad blurted out: “S___! What have we done?”
His anxiety attack didn’t last long. Within days, the trio had crafted a presentation outlining their proposed business – providing Web companies with detailed insights about their customers by combining through the mounds of user data that the businesses can’t organize themselves. After calls to a few well-placed friends and Microsoft alums, they set up their first investor meeting and snagged $5 million, more than twice the capital they initially sought.
“OBSESSED.” Among the check writes: John Cunningham of tech investor Kellett Investments and Pete Higgins, a former Microsoft group vice-president who cofounded venture-capital firm Second Avenue Partners.” Everyone walked out very jazzed about the team,” says Sam Jadallah, Internet Capital Group managing director and another Microsoft alum. “Usama has a magnetic drive that VCs love to see. (VCs love) someone obsessed with what they’re doing.”
At 37, Fayyad is certainly passionate. By all accounts, he’s one of the world’s leading brains of data-mining, a discipline that uses complex mathematical and computer tools to sniff out meaningful nuggets of information from large databases. Fayyad founded and led Microsoft’s data-mining and exploration group for nearly five years, crunching numbers for inhouse customers as well as creating data-mining applications for the company’s server software.
It was a secure, comfortable job – and that was the problem. “The one ingredient missing was risk, of having everything riding on what I was doing, and if I failed, it was going to be disaster,” he says.
The idea for Fayyad’s start-up, digiMine, sprouted after he realized that few Web businesses know how to organize the loads of information they collect online, let alone detect useful patterns. Data warehouses, which gather all of a company’s databases for easy access and analysis, were “dropping like flies,” he says, because many companies didn’t have the (information technology) support to maintain them.
BRAINIAC, digiMine’s proposition is simple: Let us do the dirty work for you. For a monthly subscription fee, the company will clean up a client’s raw data, sift through it for insights, and produce reports that business departments like marketing and sales can easily digest. Research outfit IDC forecasts that data-warehousing software and services will be a $20 billion market by 2004, up from $4.2 billion last year. Services such as digiMine’s are projected to account for 55% of that revenue.
Fayyad didn’t set out to help marketers figure out consumer behavior patterns. In fact, he preferred no career at all back in the 1980s. At the University of Michigan, he polished off bachelor’s degrees in electrical and computer engineering and master’s degrees in computer science, engineering, and mathematics.
While working on his engineering PhD, Fayyad took a summer internship at NASA’s Jet Propulsion Laboratory (JPL), a government research center that focuses on robotic exploration of the solar system. The challenge of tackling real-world projects at JPL, he says, changed his mind about being a professional student. He quickly wrapped up his thesis at age 27, headed out to join JPL’s rocket scientists.
VENUSIAN VOLCANOES. JPL offered Fayyad diverse challenges for his data-mining techniques. In one instance, he developed artificial-intelligence software tools to aid astronomers in identifying billions of objects in the night sky. Researchers discovered 16 distant quasars as a result, and Fayyad scored millions more in funding from NASA.
Another project saw him team up with geologists to pinpoint volcanoes on Venus by mining a database of some 30,000 satellite photos of the planet. Whether you’re hunting for volcanoes or start, cancerous cells or certain type of consumers, he says, the data-mining process remains pretty much the same.
By the mid ‘90s, Fayyad’s accomplishments were getting noticed by the Microsoft folks. The admiration wasn’t mutual at first – he declined their initial request for an interview. “I was very dismissive,” he says. “At JPL, we used Unix,” not a Microsoft product. But the Redmond powerhouse offered him something he couldn’t refuse: Ann all-expenses-paid vacation to Seattle for him and his family.
CHARMED. In return, it asked him to spend just one day touring the Microsoft campus. That day, he met Nick Besbeas, who was then working on market research for the MSN online service. They spoke about tracking consumer behavior online and how data mining might help. “When you talk to research scientists, you can get into philosophical questions over how things could be done,” Besbeas says. “But Usama was more practical. He came across as a guy who really knew his stuff.”
Not surprisingly, Fayyad left Seattle with a more favorable impression of Microsoft, but he claims that he still wasn’t ready to abandon JPL. Then, he says, the software king turned on its “recruiting machine.” It started sending toys to his kids and invited him to the launch party for Windows 95. One day, a delivery boy with a huge bouquet of flowers appeared at his office with two JPL security guards. Little by little, the tactics started to work. “I used to think I was immune to things like flowers and gifts,” he admits. “They shouldn’t affect your career decision, but in truth, they get to you.”
Microsoft’s commitment to long-term research also weighed heavily in Fayyad’s decision to sign up. In his first week on the job, he says, his bosses instructed him to think at least five years into the future on data-mining projects, later upping it to 10 years. Then came the clout of carrying a Microsoft business card. “If I walked into a major conference and started talking about tour plans for the next version of SQL Server, people would stop and listen,” he says. “You could push new technology and show them that data mining can play an amazing role with these database systems.”
Still, he couldn’t ignore the itch to break out on his own. Last year, he started seeking advice from old friends and colleagues, including Rajeev Motwani, a Stanford University computer-science professor and now an adviser to digiMine. “There’s a lot of ups and downs in a startup business, but Usama had the spark in him,” he says. “The person at the center of it all is what it’s all about. It’s more important than the technology.
COLLECTING RECRUITS. Meanwhile, Fayyad was also talking up his idea with Besbeas, by then a group manager in Microsoft’s application division, and Bassel Ojjeh, group program manager in its Internet business server division. Both had decided to leave the company and were thinking about what to do next. Soon the two had signed on to Fayyad’s plan.
Will their plan for digiMine fly? “The immediate value for clients is not data mining but the simple production reports that they can’t get on their own,” says Frank Gillet, senior analyst at Forrester Research. “Data mining will be gravy, if they get to it.” According to Besbeas, digiMine recently finished beta testing its product with more than a dozen companies and expects to name several charter customers for its subscription service in the coming weeks. Subscription fees start at $5,000 a month, he says.
Plenty of competitors are peddling similar services, however, IBM and SAS Institute are heavy hitters in the data-warehousing market. Web-focused outfits such as E.piphany also sell customer centric software that ferrets out who’s buying what when.
A BIG MINI-ME. Then there’s the never-ending debate over consumer privacy online. Will privacy issues cause e-businesses to limit their use of consumer data and in turn dry up business for companies such as digiMine? Fayyad says digiMine is focused on providing its clients with the infrastructure to store, manage, and put their data to use. To what degree clients plumb that data is their decision, he says, and right now, most of them are simply interested in learning how consumers navigate their Web sites.
For now, Fayyad, whom pal Jadallah calls a six-foot-four Mini Me for his resemblance to the Austin Powers character, is enjoying the ride. His one regret is that it all happened too fast. In the six months since that first breakfast at Hector’s, digiMine’s staff has grown to more than 60. And next week, the company will announce a second round of VC funding: $20 million. “In research, you can think about a problem for two months,” he says. “(At digiMine,) you just have to give it a quick thought, rely on your instincts, make a decision, an
By: Jennifer Gill
Source: Business Week Online